Skip to main content

Documentation Index

Fetch the complete documentation index at: https://uncoded.ch/docs/llms.txt

Use this file to discover all available pages before exploring further.

Multi-timeframe strategies combine longer-cycle context with faster entry timing. A daily-timeframe trend filter establishes “we’re in a tradable regime”; a 15-minute entry trigger catches the entry within that regime. Two strategies, one composed system. Better than either alone.

​
Goal

Capture entries on BTCUSDT 15m only when the daily timeframe confirms a tradable regime (uptrending). Avoid entries during sustained daily downtrends.

​
Why multi-timeframe

15m candles produce many entry signals — but many are noise that won’t follow through. Daily timeframe filters out the regime-mismatched entries.
Daily-only strategies fire infrequently. You miss entry timing opportunities within a confirmed daily trend. Adding 15m entry timing increases activity within the daily-confirmed regime.
“Daily uptrend AND 15m entry signal” is more selective than either alone. Selectivity = better per-trade quality, even at lower trade count.

​
When to pick this recipe

Crypto markets in mixed regimes

When daily timeframes show uptrends but with intra-day chop. The recipe captures the daily trend while tolerating intra-day volatility.

You're an experienced operator

Multi-strategy composition is more complex than single-strategy patterns. Best after you’ve validated single-strategy recipes.

You want a more sophisticated strategy

Beyond starter recipes, multi-timeframe is a “level-2” pattern — more involved, more powerful when validated.

​
When NOT to pick this recipe

  • ❌ First-month operators — start with single-strategy recipes.
  • ❌ You haven’t run RSI mean-reversion or EMA-cross successfully — multi-timeframe builds on those patterns.
  • ❌ Pure sideways markets — daily filter keeps canBuy: false most of the time.
  • ❌ Capital below $10,000 — composition pattern needs sufficient capital for meaningful per-trade P&L.

​
Setup steps — two strategies in composition

1

Strategy A: Daily trend filter

SignalEditor → New Strategy. Name: daily-trend-filter-btcusdt.
  • Price node: BTCUSDT, timeframe 1d.
  • Indicator: EMA(50) on price.
  • Condition: price > EMA(50) AND EMA(50) is rising over 5 daily bars (use the rising trend condition, length 5).
  • Trigger: once_per_bar_close.
  • Webhook: write trend_filter_btcusdt: true to a custom config flag.
2

Strategy A mirror — daily filter OFF

SignalEditor → New Strategy. Name: daily-trend-filter-btcusdt-off.
  • Same Price + EMA setup.
  • Condition: price < EMA(50) OR EMA(50) is falling over 5 daily bars.
  • Trigger: once_per_bar_close.
  • Webhook: write trend_filter_btcusdt: false.
3

Strategy B: 15m entry trigger

SignalEditor → New Strategy. Name: entry-trigger-btcusdt-15m.
  • Price node: BTCUSDT, timeframe 15m.
  • Indicator: RSI(14) on price.
  • Condition: RSI < 30 AND trend_filter_btcusdt == true (read the flag set by Strategy A).
  • Trigger: once_per_bar_close.
  • Webhook: write canBuy: true for the mode.
4

Strategy B mirror — entry OFF

SignalEditor → New Strategy. Name: entry-trigger-btcusdt-15m-off.
  • Same Price + RSI.
  • Condition: RSI > 70 OR trend_filter_btcusdt == false.
  • Trigger: once_per_bar_close.
  • Webhook: write canBuy: false.
5

Test the chain end-to-end

Manually drive the chain by sending test webhooks. Verify:
  • When daily trend is on AND 15m entry triggers, mode’s canBuy flips to true.
  • When daily trend turns off, mode’s canBuy flips to false regardless of 15m signals.
  • Database surface (Dashboard) shows the expected configuration state after each step.
6

Backtest the combined system

Run the Backtester with the mode + 4-strategy composition. Validate the combined behavior on multiple historical windows.
7

Forward-test on small capital

$3,000–$5,000 allocation for 4+ weeks (multi-timeframe recipes have less frequent entries — need longer windows to validate).
8

Promote if validated

Scale to full capital after forward-testing agrees with backtest.

​
Expected behavior

Entries fire only when both filters agree. Expect substantially fewer entries than RSI mean-reversion alone — perhaps 1/3 to 1/5 of the rate. The trade-off: each entry has higher selectivity.
Multi-timeframe filtering typically improves win rate over single-strategy approaches. Expect ~70–85% win rate on validated combinations.However: total return depends on win count × avg win size. Lower frequency means total return may be similar or modestly different from less-selective approaches.
Multi-timeframe with daily trend filter typically has shallower drawdowns than unfiltered approaches — the filter excludes most “buying into a downtrend” scenarios.

​
Common mistakes

Without Strategy A’s daily filter writing trend_filter_btcusdt: true/false, Strategy B’s read of the flag is undefined. Behavior collapses to RSI-only.Fix: build all 4 strategies (A, A-off, B, B-off) before going live.
Without the mirror strategies, flags stay set forever once flipped.Fix: every “set to true” strategy needs a paired “set to false” strategy. Always.
Daily trend filter with length 5 for rising is reasonable. With length 1 (just one bar’s direction), the filter trips constantly.Fix: use trend-detection lengths that match the timeframe’s character (5–20 for daily; 20–50 for hourly).
Multi-timeframe strategies have less frequent firing. A 1-week forward test may include zero entries.Fix: forward-test for at least 4 weeks; longer if your daily filter excludes most of the recent regime.

​
What’s next

RSI Mean-Reversion

The single-strategy recipe this builds on.

EMA Cross Trend

Trend-following alternative.

SignalEditor

Build the multi-strategy composition.

Backtester

Validate the composed system on history.
Last modified on May 3, 2026